| 摘要 |
PROBLEM TO BE SOLVED: To make it possible to correct speech recognition easily and speedily at the time of mis-recognition by providing candidates for recognition. SOLUTION: When a speaker inputs a speech through a microphone 3, the speech is recognized in a speech processing part 5, and the device operates based on the recognition result. When the speaker notices the mis-recognition, he or she points it out to the device by an operation like a long push on an utterance SW8. The recognition result of the inputted speech is stored in memory 9, and is displayed as a character string on a display device 2 by the indication of the mis-recognition. When the speaker looks at the display screen and selects the mis-recognized word part, words for recognition candidates are displayed, and if there is a correct word in the candidates the recognition result can be corrected. Guidance by speech instead of the display of the recognition result or both are also possible. It is also possible to perform the operation and the indication by speech. |
